/animals_classification

Primary LanguageJupyter NotebookMIT LicenseMIT

animals_classification

Markdown Monster icon

Description

(to add)

Click to see the full challenge info

Experiments pipeline

  • Hardware stack

    • RAM : 12 GB
    • Accelerator type : Nvidia GPU Geforce GTX 1060 Max-Q
    • VRAM : 6.1 GB
    • num workers : 4 (CPU count)
  • Software stack

    • Language : Python (version 3.8.6)
    • DL library :
      • Pytorch (version 1.8.1) + Pytorch Lightning (1.2.0)

The scoring metric is Accuracy

\begin{equation}
  Accuracy = \frac{ number of correct predictions }{ total number of samples }
\end{equation}

(number of correct samples / number total samples) .

Exp 1 : resnets

resnet18 (5-folds)

  • score : 0.0

Usage

  • Training
  • Inference

Results

  • What worked:
  • What didn't worked:

Acknowledgements

  • The code is based on learning from the shared notebooks on internet
  • Some of the snippets code copied from anywhere will liked to their source (original implementation)